Mechatronics 21 (2011) 109115

Contents lists available at ScienceDirect

Mechatronics
journal homepage: www.elsevier.com/locate/mechatronics

Position control of high performance hydrostatic actuation system

using a simple adaptive control (SAC) method
S.H. Cho a,, R. Burton b
a
b

Department of Mechanical and System Design Engineering, Hong-Ik University, Seoul, Republic of Korea
Department of Mechanical Engineering, University of Saskatchewan, Saskatoon, Canada

a r t i c l e

i n f o

Article history:
Received 13 February 2010
Accepted 10 September 2010
Available online 13 October 2010
Keywords:
Position control
Electro hydraulic actuator
Parallel feedforward compensator
Energy-efciency
Simple adaptive control

a b s t r a c t
This paper deals with the issue of position tracking control of a high performance hydrostatic actuation
system using simple adaptive control. For energy-efciency and savings, a speed-controlled xed displacement pump is utilized to drive a symmetrical linear actuator instead of a directional control servo
valve. The whole control system is composed of a pair of interconnected subsystems, that is, a feedback
control system and a feedforward control system to enhance the tracking performance. The experiment
using the proposed control scheme has been performed and a signicant reduction in position tracking
error is achieved compared to a conventional PID control.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Electro hydrostatic transmissions [13] have been the focus of
much interest in the uid power world due to issues associated
with energy-efciency and savings as well as environmental problems arising from CO2 emission. Variable transmission of power
can readily be achieved by directly connecting an input shaft
speed-controlled pump to a hydraulic actuator. A directional
control servo valve is not required in this scheme and thus the
power loss across it can be avoided. Most hydrostatic circuits use
a variable displacement pump driven at a xed shaft speed. The
movement of the hydraulic actuator is regulated by moving a
swash plate on the pump that changes the magnitude and direction of ow. These circuits can be energy inefcient, as the pump
continuously runs irrespective of the motion or dwell period of
the actuator. Furthermore, use of variable displacement pumps
can result in ripple effects that could degrade high precision motion [2]. The electro hydraulic actuator (henceforth referred to as
the EHA) [3,4] examined in this paper uses a xed displacement
pump whose speed and direction is controlled by an AC motor.
Hydraulic servo systems may be required to perform under a
variety of operating conditions: therefore robust control performance is important, particularly in motion control applications.
When the desired position is time varying, classical feedback control alone is not sufcient to ensure good tracking behavior. Performance can be signicantly improved through the use of a
Corresponding author. Tel.: +82 2 320 1682; fax: +82 2 322 7003.
E-mail address: shcho@wow.hongik.ac.kr (S.H. Cho).
0957-4158/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechatronics.2010.09.003

feedforward tracking controller [5]. However, parameter variations

and/or disturbances may result in mismatches between the real
plant and the model used for controller design, which may lead
to signicant degradation in tracking performance or even instability of the overall system. In such cases, adaptive algorithms have
been demonstrated to provide accurate and stable tracking performance [57]. In order to effectively adjust the parameters related
with an adaptive algorithm, a set of assumptions that form the
basis of adaptive control theory are necessary, i.e., linearity of plant
model, plant order being known, minimum phase system, and no
disturbance condition. Simple adaptive control (hence forth
referred to as SAC) is a control methodology based on a two-degree-of-freedom control system, ensuring stability of the overall
system through output feedback, while guaranteeing a tracking
capability of reference input through feedforward control methods
[7,8]. The controller structure of the SAC is simpler than the conventional adaptive control method and also it is known to be excellent in terms of control performance and robustness [9,10].
At the University of Saskatchewan, a high precision position
EHA systems has been developed by Habibi et al. [4]. This system
employs a velocity feedback inner loop controller around the DC
motor and an outer loop position controller which has resulted
in positional accuracies of 150 nm. Several control strategies has
been applied to achieve this accuracy such as nonlinear gain scheduling [12], and fuzzy controls [13], Studies to investigate tracking
performance involved the application of a variable structure lters
and sliding mode control [14] and robust discrete-time sliding
mode control [15] which yielded reasonable performance. However, it was of interest to consider an adaptive type controller to

110

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

Nomenclature
KAC
TAC
Voac
A
b
m
CLe
QLi
Qa, Qb
Q1, Q2
y

xp
a

gain of AC motor
time constant of AC motor
pipe plus mean actuator chamber volume
pressure area in actuator
viscous damping coefcient
mass load
external leakage coefcient
pump cross-port leakage coefcient
pump ow
actuator ow
displacement of actuator

d
ke(t)
kx(t)
ku(t)
kI(t)
kP(t)
ey(t)
xm(t)

investigate if the tracking performance could be improved and

hence this was the motivation of applying the SAC to this system.
The format of the paper is as follows. Section 2 describes the
EHA system used. The design process of the SAC is given in
Section 3. Experimental results with discussion are given in
Section 4. Finally, the conclusion is described in Section 5.

pump angular velocity

PFC parameter
PFC parameter
feedback gain of output error
feedback gain of reference model
feedback gain of reference input
integral gain of parameter adjusting law
proportional gain of parameter adjusting law
output error
state vector of reference model

modelled as rst-order system. Ideal working conditions are

assumed in that the supply pressure is constant, cavitation is
ignored, and the friction coefcient at any operating point is
constant. A block diagram of the equations is given in Fig. 3. The
control paths to be discussed are also shown.
 Pump ow-pressure:

Va
be
Vb
Q b Dp xp  C Li pa  pb
be

2. Electro hydrostatic actuation system

Q a Dp xp  C Li pa  pb 

Fig. 1 shows the EHA circuit [3] that forms the basis in this
study. It consists of a pair of modules, i.e., the supply module
and the actuation module. A schematic of the single rod symmetrical actuator used is illustrated in Fig. 2. There are two working
chambers C1 and C2. The surface area of the closed end of the
cylinder, A1, is made equal to the area A2 of the disc. A1 and
A2 are the active area of the two pressure chambers and are
therefore made equal (A1 = A2 = A). Chamber C3 can be opened
to the atmosphere or could be lled with pressurized gas or uid
in applications, where a bias is required to counter balance a dead
weight acting under gravity. Component specications are listed
in Table 1.
In the application of the SAC, the relative degree of the plant
model and the approximate value of the coefcient of the highest-order term in the numerator polynomial are required. In the
following section. A mathematical model is derived, which is used
in as simulation mode to evaluate the control performance. The
key equations are summarized in Eqs. (1)-(6) and are consistent
with those used in other studies [3,4,12,14]. The AC motor has been

dpa
dt
dpb
dt

1
2

 Actuator ow-pressure:

V oac Ay
be
V oac  Ay
Q 2 Ay_ 
be

Q 1 Ay_

dp1
C Le p1
dt
dp2
 C Le p2
dt

3
4

 Load ow equation:

QL

Q1 Q2 Qa Qb

2
2

 Load force equation:

Fig. 1. EHA circuit.

by_
F p1  p2 A my

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

111

(7) and (8). In order to verify the linear plant model in Fig. 4, a series of experiments have been performed for the closed-loop response and open-loop response. Fig. 5a compares the
experimental closed-loop response with that of the simulation
using the linear plant model in Fig. 4. Fig. 5b compares the experimental open-loop response with that of the simulation using the
linear plant model in Fig. 4. Ripples in the velocity waveform
may be caused by the nite number of gears in the gear pump. It
was concluded that as a rst approximation, the linear model of
the plant did represent its physical counterpart about the operating points.

v
!
u 2
u A C Le b=2 C Li b
t
be
xn
mV o


C Le m=2 C Li m bV o =be
f
be
2xn mV o

7
8

The natural frequency and damping ratio used for simulation

were determined to be: xn = 157 rad/s, f = 0.175 [3]. The open loop
gain was determined to be: K = 0.0137 ((m/s)/V). All values were
obtained experimentally through a series of repeatable open-loop
tests. The parameter values of the linear plant model are required
for controller design. In the design of the SAC, the knowledge of
plant structure is also necessary.

Fig. 2. Symmetrical actuator.

Table 1
Component specications.
Component

Parameters

Value

Symmetrical actuator

Working area
Stroke

5.05  104 m2
0.12 m

Hydraulic pump

Volumetric displacement
Maximum working pressure

1.68  107 m3/rad

20,700 kPa

Asynchronous AC motor

Gain
Time constant

1 Nm/A
0.008 s

Position sensor
Mass load

Accuracy
Weight

50  109 m
20 kg

The hydraulic capacity CH is determined by the total mean

volume Vo and effective bulk modulus be. The parameter CL = CLi + CLe/2 represents leakage in the pump-actuator system. By linearizing the nonlinear terms about an operating, a simplied model is
possible and is shown in Fig. 4. For simplicity, the AC motor has
been modelled as a proportional element, because the time constant of the AC motor is sufciently small (less than 1/100 times)
compared to that of hydraulic actuator. The natural frequency
and damping ratio for the hydraulic portion are dened in Eqs.

3. Design of the SAC system

In the design of the SAC, a reference model is required to generate the desired output signal for position tracking control. The control objective is to make the output, y, of the EHA system to track
the output, ym, of the reference model. Fig. 6 shows the basic conguration of the SAC to accomplish this objective. Gm(s) represents a
transfer function of the reference model and um is the input of reference model. The order of reference model does not depend on the
order of plant model. Considering the design specication of the
EHA for optimized motion control [3], a second-order reference
model is adopted as Eq. (9). The output error is obtained by subtracting the reference model output from the plant output. The variable adaptive gains, ke(t), kx(t), and ku(t) are adjusted by adaptive
estimation laws. The control input of the SAC system and the adaptive estimation law [7] are given by Eq. (10).

Gm s

1
0:1s 12

Fig. 3. Basic feedback control structure.

112

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

Plant Model

P-control
ym

ey

ke

K n2
2
s ( s + 2n s + n2 )

ku

y
xm
um

kx

ym

Gm(s)

ey

+
+

ke

Fig. 4. Simplied block diagram.

Gpfc (s)

0.025
0.02
Position [m]

G(s)

Fig. 6. Block diagram of the SAC with PFC.

0.015
0.01
0.6

Ref.
Exp.
Sim.

0.005
0
0.5

1.5
Time [s]

0.5

ASPR

0.4

2.5

0.3
0.2

(a) Closed-loop response (ke = 480)

0.1

0.06
0.05
Velocity [m/s]

0
1.5

Exp.
Sim.

non-ASPR

0.04
0.03

1
0.5

10

Fig. 7. Discriminate plane of ASPR condition.

0.02
0.01
0

0.1

0.2
Time [s]

0.3

0.4

(b) Open-loop response (u=2.0 volt)

Fig. 5. Validation of linear plant model through experiments.

ut ktT zt
kt ke tkx tku tT
zt ey txm tum tT
kt kP t kI t
k_ I t CI ztey t  rtkI t
kP t CP ztey t
2

r1 ey t
rt 1e
2 t r2
y

CI CTI > 0;

r1 ; r2 > 0

CP CTP > 0

9
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
;

10

The parallel feedforward compensator (PFC) is required to make

the plant Almost Strictly Positive Real (ASPR). ASPR is the property that the closed-loop system with constant output feedback
gain becomes Strictly Positive Real (SPR). For an SISO system to
be ASPR, the following three conditions have to be satised [11]:

(C1) The relative degree is 0 or 1.

(C2) The coefcient of the highest-order term in the numerator
polynomial is positive.
(C3) The system is minimum phase.
When the plant satises the ASPR condition, the SAC system
may be realizable. Generally, only a few plants satisfy the ASPR
condition. As can be seen from the transfer function of plant model
in Fig. 4, the relative degree of the plant model is three and the
condition C1 is not satised. By adding a PFC appropriately selected
to the plant as shown in Fig. 6, the augmented plant,
Ga(s) = G(s) + Gpfc(s), can be ASPR. The transfer function of the PFC
suited for a plant having the relative degree of three can be
described as Eq. (11) and the related parameters may be written
as Eq. (12) [7]. Here, the values of K and xn may be uncertain or
varied during the operation of the EHA system. So it may not be
guaranteed that Eq. (12) corresponds to the optimal choice for
the PFC parameters.
In this investigation, the transfer function of the PFC is given
using a pair of adjustable parameters, a and d, as shown in Eq.
(13), which satises the ASPR condition C1 and C2. By selecting
one parameter d in Gpfc(s), both the ASPRness of Ga(s) and the
decrease of the branch effect through Gpfc(s) can be achieved simultaneously. For example, if the range of parameter variation in the
plant model is known, it is always possible to guarantee the ASPRness of Ga(s) for a smaller value of d than arbitrarily existing value

113

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

Table 2
Design parameters for controller.
SAC

0.5

CI diag100 10 1:0 10

0.4

CP diag1200 10 10 10
r1 0:001; r2 0:01

ASPR
0.3

d 0:01

non-ASPR

0.2

PIDGPID s kP kI =s kD s
0.1
0

kP 1:2  103
2

kI 1:0  104
kD 0:02

Ref.
SAC

0.02
0.01

=4
n

0
0.5

Position [m]

of d0. By selecting the parameter value of d to be small within the

range, the branch effect may be decreased.
By applying the Hurwitzs stability criterion to the numerator
polynomial of the augmented plant, Ga(s), a 3D discriminate plane
satisfying the condition C3 can be developed as shown in Fig. 7. It is
noticed that the region satisfying the ASPRness increases when the
damping ratio is greater than 0.3. As the value of a/xn increases,
the region of non-ASPR increases approximately proportionally.
Fig. 8 shows the relationship between damping ratio f and a/xn
with the constant d(=0.01). The three small circles in Fig. 8 represent regions of constant damping (f = 0.175). Based on these observations, the ASPRness has been investigated when the damping
ratio is constant as shown in Fig. 8. For comparison, the experiment
of position tracking control has been performed for the three cases
in Fig. 8.

Position [m]

Fig. 8. ASPR condition with constant damping.

1.5

3.5

3.5

Ref.
SAC

0.02
0.01

=6
n

0
0.5

1.5

2
Time [s]

2.5

Fig. 10. Responses under non-ASPR condition.

PRESSURE
TRANSDUCER
SYMMETRICAL
ACTUATOR

MASS LOAD

2.5

POSITION SENSOR

HYDRAULIC PUMP

ACCUMULATOR

AC MOTOR

Fig. 9. Overview of experimental apparatus.

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

Position [m]

114

Ref.
SAC
PID

0.02
0.01

Position error [m]

x 10

1.5

0
2.1

2.2

Time [s]

2.3

2.4

2.5

Fig. 11. Comparison of SAC with conventional PID.

Position [m]

0.04
0.03
0.02

Ref.
SAC
PID

0.01
0
0.5

1.5

2.5

3.5

4.5

5.5

Position error [m]

-6

x 10

SAC
PID

1
0
-1

T 1 T 2 x1
n

ds 11a

12
13

s a2

4. Experimental results and discussion

SAC
PID

k2 0:01K;

11

2.5

-7

10

k1
k2

1 T 1 s1 T 2 s 1 T 2 s

k1 0:01K;

small oscillations

Gpfc s

0
0.5
20

Gpfc s

5.1

5.2

5.3

5.4

5.5

Time [s]
Fig. 12. Behaviour with varying-amplitude in reference input signal.

Position [m]

Ref.
SAC
PID

-0.01
-0.02

0.5
Disturbance [N/m]

The experimental system consisted of a symmetrical actuator,

hydraulic pump, AC motor, and mass load as shown in Fig. 9. To
achieve an effective and exible implementation of the control
scheme, a Personal Computer incorporating the National Instrument data acquisition board (NI: AT-MIO-16H-9) was employed
for real time processing of Matlab-Simulink. For experimental
work the sampling time chosen was t = 0.001 s, which was considered to be appropriate in the context of the bandwidth of the AC
motor used and the required closed-loop system dynamics. The
advantage of using speed control of the AC motor, is that it is possible to realize different kinds of control tasks in terms of energysaving. For feedback control, the piston position was measured
using a high accuracy linear encoder (resolution: 50 nm).
Design parameters for the SAC and a PID controller are listed in
Table 2. The design parameters of a PID control are derived by
assigning the dominant closed-loop pole to be 8.37 j 8.45. The
experimental results shows that the position tracking performance
is superior when the ASPRness is satised compared to when it is
not satised (see Figs. 10 and 11). Fig. 10 shows the responses
when a/xn = 4 and a/xn = 6, which corresponds to the boundary
between the ASPR and non-ASPR and the region of non-ASPR. It
is evident that both tracking performances deteriorate signicantly. It should also be noted that in the transient-state response
around t = 1.5 s, the amplitude of oscillation when a/xn = 4 is small
in comparison to that of when a/xn = 6. For both cases, the experimental apparatus as well as the actuator itself oscillated.
Fig. 11 shows the tracking performance of the SAC compared
with the conventional PID controller under the condition of
a/xn = 2. The transient response has been improved through the

1.5

2.5

3.5

4.5

5.5

300
SAC
PID

200
100
0
0.5

1.5

2.5

3.5

4.5

5.5

Position error [m]

-5

x 10
0
-2
-4
-6

SAC
PID

2.5

Time [s]

Fig. 13. Responses under spring-loaded external disturbance.

3.5

S.H. Cho, R. Burton / Mechatronics 21 (2011) 109115

SAC, albeit at the expense of a small oscillation as the tracking

approaches steady state conditions. The position error in Fig. 11
represents the tracking performance under steady-state. The steady state error at t = 2.5 s, for both the SAC and the PID controller
were 1.0  107 m and 7.0  107 m respectively. The SAC realizes a high performance position tracking about seven times higher
than its PID counterpart.
Fig. 12 demonstrates the behavior when the amplitude of the
reference signal is halved in sequence. The second plot shows
how the steady state error converges to a small error for time
>5 s. The electro-hydraulic servo system has an input amplitudedependent nonlinearity. To illustrate how the SAC performs under
these conditions, a step-wise decreasing input has been introduced
to test the robust tracking performance. Also in order to test the
tracking performance for a time-varying signal, the input signal
was modied to a changing smooth-part with an appropriate
time constant. The tracking behavior for this input has been shown
to resemble that for a sinusoidal reference.
In terms of input amplitude dependent characteristics, the SAC
achieves a higher performance than the PID controller. Fig. 13 compares the responses under an external disturbance load, simulated
by the presence of a spring on the left-hand side of the mass load in
Fig. 9. When the mass load simply contacts the spring, the actuator
starts to move left-hand side. Then the compressed spring exerts a
force on the mass, actually disturbing the motion control. During
the period (2.03.5 s), constant reaction force is acting as external
disturbance. The third plot demonstrates the steady state error
between 2 and 3.5 s. In terms of disturbance rejection characteristics, the SAC achieves a higher performance than PID control. In
view of adaptive tracking performance, the SAC shows a better response than other algorithms that the authors have designed and
implemented in the laboratory [1215]. The authors will be engaged in future comparative studies of different control methods
on the EHA.
5. Conclusion
In this paper, the AC motor-hydraulic pump unit has been used
to directly control a symmetrical linear actuator of an EHA. By
investigating the ASPRness on a 3D-plane for the design of the
SAC system, the effect of design parameters on the SAC performance has been able to be evaluated. Also, the robust operating
range of the SAC system has been studied on a 2D-plane. Based
on these studies, a series of experiments were performed according
to the variation of a/xn to clarify the correlation between the PFC
parameter and the plant model. The tracking control experiment of
the SAC has been performed and compared with that of a conven-

115

tional PID control. It has been shown that a signicant reduction in

position tracking error has been achieved by the SAC under external disturbance load. It remains a further study to enhance the
robustness of the SAC system by estimating the PFC parameters
on-line adaptively.
Acknowledgement
The authors are greatly acknowledged to the assistance of
researchers in uid power laboratory of the University of
Saskatchewan.
References
[1] Manring ND, Luecke GR. Modeling & designing a hydrostatic transmission with
a xed-displacement motor. J. Dynam. Syst. Meas. Control. 1998;120:4550.
[2] Watton J. Fluid power systems. Englewood Cliffs, NJ: Prentice-Hall; 1989.
[3] Habibi SR, Singh G. Derivation of design requirements for optimization of a
high performance hydrostatic actuation system. Int J Fluid Power
2000;1:1127.
[4] Habibi SR, Goldenberg AA. Design of a new high performance electrohydraulic
actuator. IEEE Trans Mech 2000;5:15864.
[5] Landau ID. System identication and control design. Englewood Cliffs,
NJ: Prentice-Hall; 1990.
[6] Kaufman H, Barkana I, Sobel K. Direct adaptive control algorithms: theory and
applications. 2nd ed. Springer; 1997.
[7] Iwai Z, Mizumoto I. Robust and simple adaptive control. Int J Control
1992;55(6):145370.
[8] Tsukamoto N, Yokota S. Two-degree-of-freedom control including parallel
feedforward compensator (the effectiveness in the control of 6-link electro
hydraulic serial manipulators). Trans Jpn Fluid Power Syst Soc
2003;4(6):12633 [in Japanese].
[9] Kyouizumi K, Fujita Y, Ebihara Y. Simple adaptive control method with
automatic tuning of PFC and its application to positioning control of a
pneumatic servo system. Trans Inst Sys Control Inform Engrs
2001;14(3):1029 [in Japanese].
[10] Yanada H, Furuta K. Adaptive control of an electrohydraulic servo system
utilizing online estimate of its natural frequency. Mechatronics
2007;17:33743.
[11] Zeheb H. A sufcient condition for output feedback stabilization of uncertain
systems. IEEE Trans Automat Control 1986;AC-31(11):10557.
[12] Habibi S, Burton R, Sampson E. High precision hydrostatic actuation systems
for micro and nano manipulation of heavy loads. In: Proceedings of IMECE05,
2005 ASME international mechanical engineering congress and exposition,
Annahein, Ca, November 511, Paper No. IMECE2005-82237 (CD Rom); 2005.
10 p.
[13] E. Sampson. Fuzzy control of a electrohydraulic actuator. M.Sc. Thesis,
University of Saskatchewan, Canada, May 2005.
[14] Wang S, Burton R, Habibi, S. Application of a combined sliding mode controller
and lter (SMCF) to an electrohydraulic actuator (EHA) system. In:
Proceedings of the 3rd FPNI PhD conference. Saratosa, Florida, June 1316,
Paper No. N/A (CD Rom); 2006. 15 p.
[15] Lin Y, Shi Y., Burton R.T. Modeling and analysis of a high precision gravityloaded electro-hydraulic actuator system. In: 7th international conference on
uid power (ICFP 2009). Hangzhou, China, April 79 (CD Rom); 2009.